Assessment of narratives can be helpful in making a diagnosis of developmental language disorder (DLD). One of the things that I like about narrative assessment is that it is efficient, you can analyze the narrative at different levels (words, sentences, story). For kids who are bilingual, narrative assessment can provide a way to analyze their language when there aren’t standardized tests. Additionally, it appears that bilingual children transfer what they know about story structure from one language to another so that also makes it useful.
We published a study examining bilingual children’s performance on the Test of Narrative Development. We compared Spanish-English bilingual children with and without DLD and examined how they performed on narrative comprehension compared to narrative production. We tested kids when they were in kindergarten and then again in first grade.
Like in some of our previous studies, we matched children with DLD with typical kids who had similar linguistic experience and who were the same age. For linguistic experience, we matched on number of years of English exposure as well as on current percent exposure to English and Spanish. This matching helps us to control for differences that could be attributed to age or exposure (vs. DLD). Kids on average had about 3 years experience in English when they started the study; and they had 55% exposure to English.
The figure below is from the article in JSLHR. You can see a couple of things.
- Bilingual children with typical development scored below about 1SD below the mean in kindergarten.
- By first grade they were scoring higher relative to the normative mean (well within 1SD). Children with DLD (in the paper referred to as primary language impaired– PLI) score much lower than their language matched bilingual peers.
- So, even though both groups are below the expected monolingual mean, there is still a reliable difference the children with and without DLD (we reported in a previous paper that the sensitivity and specificity for the TNL with bilinguals is about 78% sensitivity and 86% specificity when using 6 as the cutscore (in first grade).
- We found a comprehension-production (receptive-expressive) gap for children with DLD only in kindergarten. We don’t find this for typical bilinguals.
We also did follow up analysis to look at whether context made a difference. Remember, the TNL has a story recall with no pictures; a story with a series of pictures; and a story with a single picture. Both groups did the best on the retell condition with no pictures (but, kids with DLD scored lower across all conditions. A caveat here is that it is a story about a trip to McDonalds. So, we think that children were able to draw on their experiences to retell this story. This is a context that even without a picture provided a lot of context, and bilinguals retelling the story in English were pretty successful. Both groups did much better on the comprehension questions (right at the normative mean) than on production in kindergarten. This is likely because they needed to use pretty specific language to score well. We noticed in examination of correct answers, that they were more likely to recall words in English that matched the phonological patterns of Spanish (Lisa vs. Raymond), and that they recalled familiar words (McDonalds vs. purse).
So, what does this tell us about narratives in bilinguals. First, I think it’s possible to use narrative production and comprehension to determine DLD in bilinguals– even in their L2. Bilingual children with typical development may score below that of their monolingual peers in their L2 (-1SD), but bilinguals with DLD score significantly lower (-2 SD) Second, a gap (based on standard scores which are used to equalize the two) is more commonly seen in children with DLD and could be a sign of DLD (when both are significantly below the mean). The final thing is that context matters, familiarity helps children tell better stories especially in their second language.